Transcript Bank of Italy Partial Credit Guarantee Schemes – Experiences and Lessons A joint conference by the World Bank, Rensselaer Polytechnic Institute, and the.

Bank of Italy
Partial Credit Guarantee Schemes – Experiences and Lessons
A joint conference by the World Bank, Rensselaer Polytechnic
Institute, and the Journal of Financial Stability
The World Bank, Washington DC
March 13-14, 2008
Firms as Monitors of Other Firms: Mutual Loan Guarantee
Consortia and SME Finance
Francesco Columba, Leonardo Gambacorta, Paolo Emilio Mistrulli
The usual disclaimer applies. The opinions are those of the authors only
and in no way involve the responsibility of the Bank of Italy.
Bank of Italy
Motivation
• In Europe two thirds of all jobs are provided by SMEs, this
notwithstanding the literature shows that because of their
opaqueness SMEs may encounter difficulties in accessing the
credit market.
• Information asymmetries may be partially mitigated with collateral or
relationship lending, but SMEs due to lack of collateral or of a long
credit history may still find difficult to raise external finance.
• We analyse an alternative lending technology for SMEs, Mutual
Loan Guarantee Consortia (MLGC), similar to group lending.
• With MLGC a group of SMEs with individual limited collateral are
linked by a joint responsibility. Each SME contibutes to a guarantee
fund that is used as collateral to loans granted to MLGC members.
• Italy is a telling case given the importance of MLGCs and SMEs .
• We aim to ascertain if MLGCs help to mitigate information
asymmetries for SMEs.
The World Bank, Washington DC
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
2
Bank of Italy
Outline
• MLGCs characteristics.
• Stylized facts on MLGCs.
• Effects of MLGCs on the cost of credit.
• Deeper into MLGCs: peer monitoring and external funds.
• Effects of MLGCs on the quality of credit.
• Conclusions.
The World Bank, Washington DC
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
3
Bank of Italy
Italian Mutual Loan Guarantee Consortia
characteristics
• MLGCs are registered at the Bank of Italy and are subject to
prudential regulation only after a treshold of activity.
• The capital has to be more than 0.25 mln euro and at least 20% has
to be subscribed by affiliated firms; third parties (public and alike)
may subscribe capital.
• MLGCs ease SMEs access to credit in three ways:
– post collateral drawed from guarantee funds deposited in a bank
by affiliated firms and external bodies (usually monetary funds
with a loan-to-guarantee ratio between 10 and 20);
– selecting and monitoring firms;
– negotiating collectively with banks financing conditions.
• MLGCs are grouped in 5 main federations along business sectors.
The World Bank, Washington DC
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
4
Bank of Italy
Stylized facts for Italian MLGCs
• The MLGCs are around 1,000: their activity is stronger in Northern
Italy, whereas half of the MLGCs are in the South.
• The average number of firms affiliated to a MLGC is 1,900.
• The guarantees provided by MLGCs are 8 billions euro for loans of
20 billions euro.
• 80% of guarantees are monetary, the rest are personal.
• 55% of Italian banks lent to SMEs affiliated to a MLGCs; 22% of
large and medium banks, 46% of small banks.
• Typically a MLGC has a convention with 10 banks.
The World Bank, Washington DC
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
5
Bank of Italy
Effects of MLGCs on the cost of credit
• Unique data-set from Italian Credit Register and Survey on Interest
Rates with 263,000 SMEs that had a loan in 2005; 46,000 SMEs
had a guarantee posted by a MLGC.
• Test of the effects of the posting of a MLGC guarantee on the
interest rate paid on a bank loan to a SME.
• Benchmark model.
• Robustness.
• Beyond robustness.
The World Bank, Washington DC
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
6
Bank of Italy
Benchmark model : effects of MLGC on interest rate
Bank-firm interest rate
MLGC dummy
Artisan dummy
South Dummy
Sector dummy
Loan size
Nj
rih    1MLGCi   2 Southi   3 Arti   4 Sizei    j Sect ji 
j 1
Nh
   h Ban h   5 Monoi   6Garovih   7 Gartotih   ih
h 1
Bank dummy
Real garantee dummy
Single-lending dummy
The World Bank, Washington DC
Total guarantees dummy
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
7
Bank of Italy
The dependent variable is the interest rate on overdraft loans for firms
with less than 20 employees and for artisan firms. OLS estimates with
fixed effects for economic activity sector and for lending bank. Fixed
effects are not reported. Standard errors with white correction are in
italics. *** 1 per cent significance. ** 5 per cent. * 10 per cent.
Explicative variables
Benchmark model
firm guaranteed from a MLGC (MLGC )
0.011
0.253 ***
Southern Italy firm (South )
0.016
0.031 ***
artisan firm (Art )
0.012
-0.086 ***
log of loan used (Size )
0.005
firm borrowing from only one bank (Mono )
0.373 ***
0.009
real guarantees on overdraft loan (Garov )
-1.304 ***
existence of any type of guarantee on other credit
lines (Gartot )
0.982 ***
costant ( )
adjusted R2
Number of observations
The World Bank, Washington DC
-0.181 ***
0.019
0.010
10.298 ***
2.490
0.205
347,461
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
8
Bank of Italy
Robustness and beyond
• Robustness:
–
–
–
–
Controls for firms riskiness and bank entry;
Control for banks operating with at least a MLGC;
Controls for multiple lending and firm fixed effects;
Geographical fixed effects.
• Beyond robustness:
– Cooperative banks: a MLGC guarantee raises the interest rate of the loan;
– Control for selection bias: a treatment effect model to take into account the
decision of a firm to join a MLGC.
The World Bank, Washington DC
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
9
Bank of Italy
Sample composed of cooperative banks only. The dependent variable is
the interest rate on overdraft loans for firms with less than 20 employees
and for artisan firms. OLS estimates with fixed effects for economic
activity sector and for lending bank. Fixed effects are not reported.
Standard errors with white correction are in italics. *** 1 per cent
significance. ** 5 per cent. * 10 per cent.
Explicative variables
firm guaranteed from a MLGC (MLGC )
0.165 ***
0.037
-0.448 ***
Southern Italy firm (South )
0.143
0.045
artisan firm (Art )
0.041
-0.152 ***
log of loan used (Size )
0.015
firm borrowing from only one bank (Mono )
real guarantees on overdraft loan (Garov )
0.265 ***
0.030
-1.657 ***
0.054
existence of any type of guarantee on other credit
lines (Gartot )
1.018 ***
costant ( )
8.560 ***
adjusted R2
0.303
Number of observations
The World Bank, Washington DC
Benchmark model
0.030
2.082
25,721
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
10
Bank of Italy
The dependent variable is the interest rate on overdraft loans for firms with
less than 20 employees and for artisan firms. Maximum likelihood
estimates of a treatment effects model with fixed effects for economic
activity sector and for lending bank. errors with white correction are in
italics. *** 1 per cent significance. ** 5 per cent. * 10 per cent.
Explicative variables
firm guaranteed from a MLGC (MLGC )
Southern Italy firm (South )
Benchmark equation
-0.622 ***
0.071
0.171 ***
0.018
0.081 ***
artisan firm (Art )
0.019
-0.067 ***
log of loan used (Size )
0.006
firm borrowing from only one bank (Mono )
real guarantees on overdraft loan (Garov )
existence of any type of guarantee on other credit
lines (Gartot )
0.405 ***
0.012
-1.279 ***
0.024
0.951 ***
0.012
10.191 ***
costant ( )
2.550
selection equation for MLGC
blood donations (Blood)
0.001
-0.026 ***
black economy (black)
0.001
0.569 ***
artisan firm (Art )
0.006
retail sector firm (Retail)
building sector firm (Building)
0.047 ***
0.007
-0.118 ***
0.009
1.228 ***
State support (State)
0.023
0.092 ***
Rho
0.015
Wald Chi2
39,684
Number of observations
The World Bank, Washington DC
0.006 ***
230,492
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
11
Bank of Italy
Deeper into MLGCs: peer monitoring and external
funds
• Sub-sample of firms affiliated to a MLGC.
• Model of the firm’s choice of affiliation: Heckman procedure.
• Peer monitoring: the interest rate advantage of being affiliated with
a MLGC raises up to a maximum and then declines coherently with
a priori.
• External (public or alike) funds: reduction of interest rate advantage
of the affiliation with a MLGC points to moral hazard problems.
The World Bank, Washington DC
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
12
Bank of Italy
“Optimal” number of firms in MLGC for peer
monitoring
10.0
interest rate
9.5
The benefit on the interest rate vanishes
when the number of firms in the MLGC
overcomes 17.000 units
9.0
The size effect on the interest rate is
optimal when the number of firms in the
MLGC is around 8.500
8.5
8.0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
number of firms in a MLGC (in thousands)
The World Bank, Washington DC
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
13
Bank of Italy
Effects of MLGC on the quality of credit
Pr (bad  1 )  Φ(   MLGC   South   Art 
i
0 1
i
2
i 3 i
Nj
  Size   5 Monoi   Sect )
4
i
j
ji
j 1
The World Bank, Washington DC
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
14
Bank of Italy
The dependent variable is the probability that a firm was
classified between June 2004 and June 2005 as having a bad
debt with at least one of the lending banks. Probit estimates
with fixed effects for economic activity sector. Marginal
effects computed for a discrete variation of the dummy
variables form 0 to 1. Fixed effects are not reported. Standard
errors with white correction are in italics. *** 1 per cent
significance. ** 5 per cent. * 10 per cent.
Explicative variables
firm guaranteed from a MLGC (MLGC )
Southern Italy firm (South )
-0.016 ***
0.001
0.035 ***
0.002
-0.032 ***
artisan firm (Art )
0.001
log of loan used (Size )
firm borrowing from only one bank
(Mono )
Pseudo R2
The World Bank, Washington DC
(1)
Benchmark
equation
-0.011 ***
0.001
-0.046 ***
0.001
0.113
Log-likelihood
-60,024
Number of observations
385,008
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
15
Bank of Italy
Conclusions
•
SMEs affiliated with a MLGC obtain credit at a lower interest rates
than other SMEs, particularly where asymmetric information
problems are most severe.
•
Peer monitoring is beneficial to MLGC up to a treshold.
•
External funds in MLGC might rise moral hazard problems.
•
Firms affiliated to a MLGC are ex-post less risky.
•
MLGCs seem to be a lending technology beneficial to SMEs.
The World Bank, Washington DC
Columba, Gambacorta and Mistrulli - Firms as Monitors of Other Firms
16